This is a variant annotation and evaluation pipeline as described in A comprehensive and bias-free evaluation of genomic variant clinical interpretation tools. Some scoring tools (SIFT4G, Polyphen-2, MutationTaster2021, SpliceAI) only offer command-line or web-based programs. You can follow the instruction below to download and annotate variants with those tools separately, then use provided scripts to format and combine with the annotation from precomputed databases.

To optimize the analysis, we use Dython as a submodule and make some alternations to the package. The pipeline and optimized packages can be downloaded as followed.

git clone  --recurse-submodules https://github.com/nmtrang00/Var_Annot_Eval

If you forget to use --recurse-submodules, please proceed with two lines of codes below.

git submodule init
git submodule update

The links to github or website of tools not offering precomputed data. SIFT4G: https://sift.bii.a-star.edu.sg/ Polyphen-2: http://genetics.bwh.harvard.edu/pph2/bgi.shtml MutationTaster2021: https://www.genecascade.org/MutationTaster2021/info/ SpliceAI: https://github.com/Illumina/SpliceAI.git Required python packages in requirements.txt can be installed with pip or conda. To prevent any undesired, it is recommended to setup a conda env as followed.

conda create -n bwtool python=3.6
conda activate bwtool
conda install -c bioconda htslib==1.3.2
conda install -c pwwang bwtool
conda install -c bioconda openssl=1.0
conda install -c bioconda cyvcf2==0.10.0
conda install -c bioconda pandas==1.1.5
conda install -c anaconda seaborn==0.11.1
conda install -c conda-forge matplotlib==3.4.1
conda install -c conda-forge ppscore==1.2.0
pip install git+https://github.com/mainguyenanhvu/dython.git

After annotating variants with command-line or web-based programs, do as follow to turn the output the format that can be combined with other precomputed scores.\

Include -t option when running SIFT to get annotations for all possible transcripts, else SIFT automatically chooses the annotation for the 1st transcript. Get the most severe consequences of each variant from SIFT4G:

python3 annotation_scripts/get_SIFT_most_severe.py -i [SIFTannotations.xls] -r [inVCF] -o [output]

Required parameters:\

• -i: Dir to SIFT output ended with "_SIFTannotations.xls"\
• -r: Dir to inVCF\
• -o: Dir to output TAB with 6 columns: "var", ""SIFT_transcript_id", "SIFT_variant_type","SIFT_score", "SIFT_median", "SIFT_prediction"

Download databases needed for the annotation.

./annotation_scripts/get_databases.sh [resources]

Required parameters:\

• resources: Directory to databases storage folder.
  In order to download PrimateAI datbases, you need to have an Illumina account. Please do as instructed in the following link PrimateAI and download file name "PrimateAI_scores_v0.2.tsv.gz" to the same folder storaging other databases. Preprocess databases for optimal query:

./annotation_scripts/prepare_datbases.sh [resources]

Required parameters:\

• resources: Directory to databases storage folder.

Run the following script

./annotation_scripts/run_annotation.sh [inVCF] [resources] [output] [keeping_tmp_files]

Required parameters:\

• invcf: Directory to input VCF with variants of interest.\
• resources: Directory to databases storage folder.\
• output: Directory to annotated TAB.\
• keeping_tmp_files: 0 to remove all tmp scoring files, 1 to keep all of them.

To generate ppscore and association score, use the following command:

python annotation_evaluation.py -i [inCSV] -o [outputFolder]

• -i: Directory to input tab-formated file.
• -o: Directory to a output folder

@inproceedings{trang2021comprehensive,
  title={A comprehensive and bias-free evaluation of genomic variant clinical interpretation tools},
  author={Trang, Nguyen Minh and Vu, Mai Nguyen Anh and Anh, Tran Hoang and Nguyen, Nguyen Thanh and others},
  booktitle={2021 13th International Conference on Knowledge and Systems Engineering (KSE)},
  pages={1--5},
  year={2021},
  organization={IEEE}
}